Random thought of the day
-
@Gąska Well, if it weren't going to get flattened into the signal spectrum, it would. Sadly it's gonna. Some sort of steganography seemed to be the gone-for thing.
-
@Gribnit if you're going for steganography, you might as well go for an encryption scheme that doesn't involve flashing infrared light over the scene.
-
@Gąska If I had full-spectrum record and playback, I'd be fine with using infrared steg encoding, actually. But I - just don't have that. I would be better off using an HF strobe somewhere for steg.
-
@Gribnit let's say you have equipment capable of emitting as well as recording all wavelemgths from AM to gamma rays with resolution of half dozen Planck's lengths. How are you going to use it to verify authenticity of recordings?
-
@Gąska I need to be able to find and verify a steganographic or metadata signature. If I have an audience watching only the visible, I can encode signatures in the non-visible. If the audience is watching the whole spectrum, I need to hide it outside their visual processing or in metadata. Metadata is cheaper but more f-able with, a steg signature is wanted. Like a strobe frinstance, or messing with corner brightness, or stuff like that.
The point of infrared would be, in full-spectrum, that humans don't see it, so a signature could be scribbled there.
-
@Gąska said in Random thought of the day:
@Gribnit let's say you have equipment capable of emitting as well as recording all wavelemgths from AM to gamma rays with resolution of half dozen Planck's lengths. How are you going to use it to verify authenticity of recordings?
Alt: With this equipment, I make recordings on Earth using natural light. The recordings are intrinsically verifiable to location and exact time, by correlation with solar observatories.
-
@Gribnit said in Random thought of the day:
Metadata is cheaper but more f-able with, a steg signature is wanted.
Also known as security by obscurity.
-
@Gąska well, no, not really. The signature scheme doesn't get to be any weaker for being steg encoded vs in metadata. Metadata is easily strippable, tho - the signature is more likely to survive in a robust steg encoding.
-
@Gribnit said in Random thought of the day:
@Gąska well, no, not really. The signature scheme doesn't get to be any weaker for being steg encoded vs in metadata.
Conversely, it doesn't get any harder either.
Metadata is easily strippable, tho - the signature is more likely to survive in a robust steg encoding.
It's not about removing the signature. It's about copying it over to another, counterfeit video, so it passes as genuine. Steganography doesn't do anything to prevent that at all once the attacker knows where to look.
-
@Gąska That's why add time and/or nonce, such that signature hash includes time/nonce and content. I do not see existing signing schemes losing applicability, just need to get their plumbing in place. If I produce signature with private key, of unique-fied content+time hash , I am verifiable vs replay attack.
-
@Gribnit okay, so now you add the recorded content as input for encryption algorithm. This actually can be used to make sure that video is genuine, as long as the key is strong enough and kept secure. I mean, it's done all the time in real life. There's just one problem - you don't need IR laser to do that. Digital metadata works just as well. There's literally zero benefit of using IR steganography for this.
-
@Gąska The only benefit is that any watermark scheme adds. If infrared would work, it'd make an okay watermark. Infrared won't work, but, a steg scheme adds advantage of being extremely hard to strip.
-
@Gribnit said in Random thought of the day:
@Gąska The only benefit is that any watermark scheme adds.
Watermark isn't for verifying authenticity. It's for verifying theft. Two different use cases.
-
@Gąska You have very hard categorization schemes internally. It must make it hard to make new things.
A watermark is a forced authentication.
-
@Gąska The point of the lights is that it precisely shows all the shapes and angles of everything, which is extremely difficult to replicate when making a fake video but hopefully not so hard to check.
The cryptography part was just to make the pattern impossible to predict for an attacker.
-
@Gribnit and we've reached the point where I literally can't understand what you're even talking about. Now it's not just that you're wrong - I don't even know what you're saying.
-
@anonymous234 said in Random thought of the day:
@Gąska The point of the lights is that it precisely shows all the shapes and angles of everything, which is extremely difficult to replicate when making a fake video but hopefully not so hard to check.
We're talking about people who can make deep fakes. Compared to that, your IR laser show sounds like a child's play.
The cryptography part was just to make the pattern impossible to predict for an attacker.
They don't have to predict if they have the source material at least as long.
-
@Gąska said in Random thought of the day:
@Gribnit and we've reached the point where I literally can't understand what you're even talking about. Now it's not just that you're wrong - I don't even know what you're saying.
So a watermark says, "This came from here" in a difficult to remove or repudiate way. If you want to be able to have evidence-standard video irrepudiability is handy...
-
Speedrunners are weird. Glitching the hell out of the game in ways that developers never imagined is fair game, but glitching it in the way developers intended is banned.
-
@Gąska said in Random thought of the day:
@Gribnit and we've reached the point where I literally can't understand what you're even talking about. Now it's not just that you're wrong - I don't even know what you're saying.
I see you've met @Gribnit.
-
@anonymous234 said in Random thought of the day:
I was thinking about how to record video so that you can prove it's real and not a deepfake or similar manipulated thing, for the near future (or even present?) when that will become a real issue when using security footage in court cases.
It is impossible to prove that a picture was not faked, just like it is impossible to prove that any other kind of evidence was not faked or planted. You can make records of the collection of evidence (which can be forged) and keep records and sworn statements that evidence was correctly handled from crime scene to lab to trial (more forgery of the chain of custody possible), but ultimately you have to trust the humans to follow the procedures. You can increase trust by legislating harsh penalties for forgery and perjury.
Perhaps, something like this would lend more confidence to image veracity:
- Camera takes picture.
- The pixel data is hashed and the result is written to the picture metadata. This will show that the image was not manipulated outside the camera.
- A copy of the picture is encrypted with a public key unique to that camera. The private key is kept by the camera manufacturer. This identifies the camera that took the picture.
- The encrypted file is signed by the photographer's private key. This identifies the photographer that took the picture. This should be through some attachment to the camera that is disconnected when the camera is not in the photographer's possession.
Then, at trial, the photographer and the camera can be identified, and the pixels can be verified as unchanged. But, this requires that the various keys were secured and that the camera was tamper-resistant. Plus, you could always just create a fake image and photograph it with the secure camera. This is why physical evidence at trial is always presented with human witnesses to interpret it and to authenticate it.
-
I do recall there were attempts to prevent camrips and telesync (captured by a digital cameras in cinemas before or during the first release) from being made. It would identify the place and the exact time the capture was made by encoding that information in the source. It was also designed to technomagically survive various compression methods.
https://en.wikipedia.org/wiki/Coded_anti-piracy#CineFence <--
I'm to google what's become of it. Probably ended up being a waste of money.
-
@pie_flavor said in Random thought of the day:
@Gąska said in Random thought of the day:
@Gribnit and we've reached the point where I literally can't understand what you're even talking about. Now it's not just that you're wrong - I don't even know what you're saying.
I see you've met @Gribnit.
This will make no sense, but here's the algorithm:
Starting from a painfully low amount of compression, increase compression until the channel breaks, then maybe back off.
-
@Gąska said in Random thought of the day:
you can encrypt it with nuclear missiles launch codes if you want
-
@HardwareGeek said in Random thought of the day:
@Gąska said in Random thought of the day:
you can encrypt it with nuclear missiles launch codes if you want
OTP XOR
-
I wonder if Snowflake Computing is debating about changing their name...
-
@dcon I don't think "snowflake" has infiltrated corporate speak yet.
-
Using locale settings to retrieve currency symbol for number formatting is absolutely retarded on every level and why would there even be API for that.
-
@Gąska simplest case - apps that are written to assume it, like single-currency POS.
-
@Gribnit they could've hardcoded it. It would make a whole fucking lot more sense to hardcode it than read it from locale.
-
@Gąska No, not for a single-currency POS app. It works in the local currency, per the locale. That's a meh, but at least as valid as needed, use case.
-
This post is deleted!
-
@Gribnit it would make sense, if all business and financial laws in all countries around the world were identical. Unfortunately, they aren't. Probability of POS system designed for one country being suitable for another country is between very slim and outright nonexistent. And changing currency depending on locale makes it even less likely to work, not more. That said, I'm sure some have tried exactly that. And my personal opinion is that they should be burned at the stake.
-
@Gąska said in Random thought of the day:
POS system
I keep reading that as “piece of shit system”; I don't think I'm wrong…
-
@dkf all systems are POS systems.
-
@Gąska
Except for my intelligent Japanese toilet
-
@Luhmann I'd feel bad for shitting on intelligent beings.
-
@Gąska said in Random thought of the day:
I'd feel bad for shitting on intelligent beings.
how do you survive on this forum?
-
@Luhmann I said, intelligent.
-
@Gąska
Ok ok, but outside of the garage ...
-
@Luhmann outside garage, I mostly shit on shitty developers.
-
I can't believe Android only has a simple data usage warning at X MB. What matters is not that I've used x% of my data for the month, it's that I've used x% of the data before x% of the month. Why not be just a little bit smarter and warn me when I'm just starting to use too much data, before I'm already screwed for the month?
-
I wonder what will happen when Android runs out of Alphabet to name the versions?
-
@Applied-Mediocrity said in Random thought of the day:
I wonder what will happen when Android runs out of Alphabet to name the versions?
It can't run out of Alphabet, it's owned by it
-
@TimeBandit said in Random thought of the day:
@Applied-Mediocrity said in Random thought of the day:
I wonder what will happen when Android runs out of Alphabet to name the versions?
It can't run out of Alphabet, it's owned by it
Pictures an android running out of the front door of an office building, with generic executives in pursuit.
-
@PleegWat said in Random thought of the day:
@TimeBandit said in Random thought of the day:
@Applied-Mediocrity said in Random thought of the day:
I wonder what will happen when Android runs out of Alphabet to name the versions?
It can't run out of Alphabet, it's owned by it
Pictures an android running out of the front door of an office building, with generic executives in pursuit.
That would be the "Potato" release. As in Hot Potato.
-
You know, if they made the
rm
command's--no-preserve-root
command require a machine-specific generated parameter (given in a unpredictable output, like a text-based captcha), those jokes might go away.
-
@Tsaukpaetra Require all shell scripts to be signed by a trusted authority! Microsoft does it so it must be a good idea.
-
I've heard people say "Amazon can't fully automate their warehouses because things come in all sorts of shapes that robots can't reliably grab and handle".
Which makes sense. But then why not simply refuse to sell anything that's not in a cuboid box meeting certain specifications? This would shift the handling and packaging burden for any irregularly shaped objects to the original factory, which is a good thing, since that's who's best suited to do it.
-
@anonymous234 but it doesn't make sense. If my grabamagig is a suction cup, I can suction cup up anything with a close-to-flat surface on it.